Quantuva is a research effort in quantum software. We study the layer where a quantum algorithm meets a real device — circuit transpilation, qubit mapping, decoherence, and the error mitigation that stands between a measured answer and the one you meant. We build the tooling and publish the notes, open.
Given a backend's topology and native gate set, how do you rewrite an algorithm into the shortest circuit that still computes the same thing? Gate decomposition, SWAP routing, and scheduling.
A fast local simulator for the day-to-day, and a density-matrix simulator for modelling decoherence and readout error before you spend hardware time.
Zero-noise extrapolation, Pauli twirling, and readout correction — methods that trade classical sampling budget for a less biased expectation value.
On matching the simulation step to the observable you actually measure, not the one you can compute.
2026-07-02How few noise levels you really need before extrapolation stops being worth the overhead.
2026-05-27Estimating success probability without a full reference-state reconstruction.
The SDK treats a circuit and its observables as first-class objects. You describe the physics; the transpiler, simulator, and mitigation passes handle the parts that have nothing to do with your science.
# GHZ state with readout-error mitigation from quantuva import Circuit, Observable, transpile qc = Circuit(3) qc.h(0) qc.cx(0, 1) qc.cx(0, 2) qc.measure_all() backend = transpile.target("superconducting-v2") compiled = transpile(qc, backend, opt_level=2) obs = Observable("Z(0) @ Z(1)") result = compiled.run(shots=2048, mitigate="zne") print(result.expectation(obs)) # -> a calibrated ⟨Z₀⊗Z₁⟩
Bell pairs, GHZ states, a four-qubit QFT, and a surface-code patch — each with the circuit, its statevector, and the routine that produced it.
Entangling three qubits with a single Hadamard and two CNOTs, then measuring the stabilizer.
example 02Controlled phase rotations on four qubits, and the SWAP that undoes the bit reversal.
example 03A small surface code layout with its X and Z stabilizers drawn as vertices and faces.
A qubit-mapping pass that folds SWAP insertion into the search, rather than patching it on afterward.
QV-2026-02Compiling to native two-qubit interactions with fewer entangling gates than fixed templates.
QV-2025-05A fit-free ZNE estimator and the conditions under which it degrades gracefully.
Quantuva is written in the open. Follow along, reproduce the experiments, or contribute — there is no separate academic edition.
Join the community →